基于WOA-DBN模型的支架载荷预测研究分析  

Research and Analysis of Support Load Prediction Based on WOA-DBN Model

作　　者：鲁杰[1,2] 张松 杨志强 王劭琛 魏征 刘泽 LU Jie;ZHANG Song;YANG Zhiqiang;WANG Shaochen;WEI Zheng;LIU Ze(College of Coal Engineering,Shanxi Datong University,Datong,Shanxi 037009,China;College of Mining Engineering,China University of Mining and Technology,Xuzhou,Jiangsu 221116,China;Zhongtian-hechuang Energy Co.,Ltd.,Ordos,Inner Mongolia 017004,China)

机构地区：[1]山西大同大学煤炭工程学院,山西大同市037009 [2]中国矿业大学矿业工程学院,江苏徐州市221116 [3]中天合创能源有限责任公司,内蒙古鄂尔多斯市017004

出　　处：《矿业研究与开发》2025年第3期222-228,共7页Mining Research and Development

基　　金：山西省基础研究计划项目(202303021222203);山西大同大学2023年科研创新项目(23CX54)。

摘　　要：在矿山生产中,工作面冒顶事故与液压支架直接相关。依据这一理论,提出了一种基于多源数据融合的预测模型,用于预测液压支架的载荷。通过研究煤层顶板来压变形特性理论、液压支架的组成及工作原理、承载理论以及工作面工况对液压支架的影响,分析载荷变化的影响因素,并对关键受力元件进行数据采集。采用K均值聚类算法对数据的特征进行聚类分析,对载荷进行分类预测建模。利用鲸鱼优化算法(WOA)分别优化长短时记忆网络(LSTM)和深度信念神经网络(DBN),建立WOA-LSTM串联式预测模型和WOA-DBN串联式预测模型。结果表明,WOA-DBN模型在对20^(#)液压支架前立柱载荷预测中,平均绝对误差分别降低了0.2287,0.2064,0.0677;均方根误差分别降低了0.2129,0.1953,0.0725。WOA-DBN模型对20^(#)液压支架后立柱载荷预测中,平均绝对误差分别降低了0.3031,0.2446,0.2054;均方根误差分别降低了0.2919,0.2464,0.2389。可见,WOA-DBN串联式预测模型更适合载荷预测且精度更高。In mine production,the roof fall accident of working face is directly related to the hydraulic support.According to this theory,a prediction model based on multi-source data fusion was proposed to predict the load of hydraulic support.By studying the theory of pressure deformation characteristics of coal seam roof,the composition and working principle of hydraulic support,the bearing theory and the influence of working condition on hydraulic support,the influencing factors of load change were analyzed,and the data of key stress components were collected.The K-means clustering algorithm was used to cluster the characteristics of the data,and the load was classified and predicted.The whale optimization algorithm(WOA)was used to optimize the long short-term memory network(LSTM)and the deep belief neural network(DBN)respectively,and the WOA-LSTM series prediction model and the WOA-DBN series prediction model were established.The results are as follows.Firstly,the mean absolute error is reduced by 0.2287,0.2064 and 0.0677 respectively in the prediction of the front column load of 20^(#)hydraulic support by WOA-DBN model.The root mean square error is reduced by 0.2129,0.1953 and 0.0725,respectively.Secondly,in the prediction of the rear column load of 20^(#)hydraulic support by WOA-DBN model,the mean absolute error is reduced by 0.3031,0.2446 and 0.2054 respectively.The root mean square error is reduced by 0.2919,0.2464 and 0.2389,respectively.It can be seen that the WOA-DBN series prediction model is more suitable for load prediction and has higher accuracy.

关 键 词：支架载荷预测 多源数据融合 WOA-DBN K均值聚类算法 

分 类 号：TD353[矿业工程—矿井建设]